def get_dataset(device):
# with open(params.api_dir, "rb") as fh:
# api = pkl.load(fh, encoding='latin1')
api = SWDADialogCorpus(params.data_dir)
dial_corpus = api.get_dialog_corpus()
train_dial, test_dial = dial_corpus.get("train"), dial_corpus.get("test")
# convert to numeric input outputs
train_loader = SWDADataLoader("Train",
train_dial,
params.max_utt_len,
params.max_dialog_len,
device=device)
valid_loader = test_loader = SWDADataLoader("Test",
test_dial,
params.max_utt_len,
params.max_dialog_len,
device=device)
if api.word2vec is not None:
return train_loader, valid_loader, test_loader, np.array(api.word2vec)
else:
return train_loader, valid_loader, test_loader, None
def main():
# config for training
config = Config()
config.use_bow = False
# config for validation
valid_config = Config()
valid_config.keep_prob = 1.0
valid_config.dec_keep_prob = 1.0
valid_config.batch_size = 60
valid_config.use_bow = False
# configuration for testing
test_config = Config()
test_config.keep_prob = 1.0
test_config.dec_keep_prob = 1.0
test_config.batch_size = 1
test_config.use_bow = False
pp(config)
# get data set
api = SWDADialogCorpus(FLAGS.data_dir,
word2vec=FLAGS.word2vec_path,
word2vec_dim=config.embed_size)
dial_corpus = api.get_dialog_corpus()
meta_corpus = api.get_meta_corpus()
train_meta, valid_meta, test_meta = meta_corpus.get(
"train"), meta_corpus.get("valid"), meta_corpus.get("test")
train_dial, valid_dial, test_dial = dial_corpus.get(
"train"), dial_corpus.get("valid"), dial_corpus.get("test")
# convert to numeric input outputs that fits into TF models
train_feed = SWDADataLoader("Train", train_dial, train_meta, config)
valid_feed = SWDADataLoader("Valid", valid_dial, valid_meta, config)
test_feed = SWDADataLoader("Test", test_dial, test_meta, config)
# begin training
sess_config = tf.ConfigProto(log_device_placement=False,
allow_soft_placement=True)
# sess_config.gpu_options.allow_growth = True
sess_config.gpu_options.per_process_gpu_memory_fraction = 0.45
with tf.Session(config=sess_config) as sess:
initializer = tf.random_uniform_initializer(-1.0 * config.init_w,
config.init_w)
scope = "model"
with tf.variable_scope(scope, reuse=None, initializer=initializer):
model = KgRnnCVAE(sess,
config,
api,
log_dir=None if FLAGS.forward_only else log_dir,
forward=False,
scope=scope)
with tf.variable_scope(scope, reuse=True, initializer=initializer):
valid_model = KgRnnCVAE(sess,
valid_config,
api,
log_dir=None,
forward=False,
scope=scope)
with tf.variable_scope(scope, reuse=True, initializer=initializer):
test_model = KgRnnCVAE(sess,
test_config,
api,
log_dir=None,
forward=True,
scope=scope)
test_model.prepare_mul_ref()
logger.info("Created computation graphs")
if api.word2vec is not None and not FLAGS.forward_only:
logger.info("Loaded word2vec")
sess.run(model.embedding.assign(np.array(api.word2vec)))
# write config to a file for logging
if not FLAGS.forward_only:
with open(os.path.join(log_dir, "run.log"), "wb") as f:
f.write(pp(config, output=False))
# create a folder by force
ckp_dir = os.path.join(log_dir, "checkpoints")
if not os.path.exists(ckp_dir):
os.mkdir(ckp_dir)
ckpt = tf.train.get_checkpoint_state(ckp_dir)
logger.info("Created models with fresh parameters.")
sess.run(tf.global_variables_initializer())
if ckpt:
logger.info("Reading dm models parameters from %s" %
ckpt.model_checkpoint_path)
model.saver.restore(sess, ckpt.model_checkpoint_path)
if not FLAGS.forward_only:
dm_checkpoint_path = os.path.join(
ckp_dir, model.__class__.__name__ + ".ckpt")
global_t = 1
patience = 10 # wait for at least 10 epoch before stop
dev_loss_threshold = np.inf
best_dev_loss = np.inf
for epoch in range(config.max_epoch):
logger.info(">> Epoch %d with lr %f" %
(epoch,
sess.run(model.learning_rate_cyc,
{model.global_t: global_t})))
# begin training
if train_feed.num_batch is None or train_feed.ptr >= train_feed.num_batch:
train_feed.epoch_init(config.batch_size,
config.backward_size,
config.step_size,
shuffle=True)
global_t, train_loss = model.train(
global_t,
sess,
train_feed,
update_limit=config.update_limit)
# begin validation
logger.record_tabular("Epoch", epoch)
logger.record_tabular("Mode", "Val")
valid_feed.epoch_init(valid_config.batch_size,
valid_config.backward_size,
valid_config.step_size,
shuffle=False,
intra_shuffle=False)
valid_loss = valid_model.valid("ELBO_VALID", sess, valid_feed)
logger.record_tabular("Epoch", epoch)
logger.record_tabular("Mode", "Test")
test_feed.epoch_init(valid_config.batch_size,
valid_config.backward_size,
valid_config.step_size,
shuffle=False,
intra_shuffle=False)
valid_model.valid("ELBO_TEST", sess, test_feed)
# test_feed.epoch_init(test_config.batch_size, test_config.backward_size,
# test_config.step_size, shuffle=True, intra_shuffle=False)
# test_model.test_mul_ref(sess, test_feed, num_batch=5)
done_epoch = epoch + 1
# only save a models if the dev loss is smaller
# Decrease learning rate if no improvement was seen over last 3 times.
if config.op == "sgd" and done_epoch > config.lr_hold:
sess.run(model.learning_rate_decay_op)
if valid_loss < best_dev_loss:
if valid_loss <= dev_loss_threshold * config.improve_threshold:
patience = max(patience,
done_epoch * config.patient_increase)
dev_loss_threshold = valid_loss
# still save the best train model
if FLAGS.save_model:
logger.info("Save model!!")
model.saver.save(sess, dm_checkpoint_path)
best_dev_loss = valid_loss
if (epoch % 3) == 2:
tmp_model_path = os.path.join(
ckp_dir,
model.__class__.__name__ + str(epoch) + ".ckpt")
model.saver.save(sess, tmp_model_path)
if config.early_stop and patience <= done_epoch:
logger.info("!!Early stop due to run out of patience!!")
break
logger.info("Best validation loss %f" % best_dev_loss)
logger.info("Done training")
else:
# begin validation
# begin validation
valid_feed.epoch_init(valid_config.batch_size,
valid_config.backward_size,
valid_config.step_size,
shuffle=False,
intra_shuffle=False)
valid_model.valid("ELBO_VALID", sess, valid_feed)
test_feed.epoch_init(valid_config.batch_size,
valid_config.backward_size,
valid_config.step_size,
shuffle=False,
intra_shuffle=False)
valid_model.valid("ELBO_TEST", sess, test_feed)
dest_f = open(os.path.join(log_dir, "test.txt"), "wb")
test_feed.epoch_init(test_config.batch_size,
test_config.backward_size,
test_config.step_size,
shuffle=False,
intra_shuffle=False)
test_model.test_mul_ref(sess,
test_feed,
num_batch=None,
repeat=5,
dest=dest_f)
dest_f.close()
def main():
# config for training
config = Config()
print("Normal train config:")
pp(config)
valid_config = Config()
valid_config.dropout = 0
valid_config.batch_size = 20
# config for test
test_config = Config()
test_config.dropout = 0
test_config.batch_size = 1
with_sentiment = config.with_sentiment
###############################################################################
# Load data
###############################################################################
# sentiment data path: ../ final_data / poem_with_sentiment.txt
# 该path必须命令行显示输入LoadPoem，因为defaultNonehjk
# 处理pretrain数据和完整诗歌数据
# api = LoadPoem(args.train_data_dir, args.test_data_dir, args.max_vocab_size)
api = LoadPoem(corpus_path=args.train_data_dir,
test_path=args.test_data_dir,
max_vocab_cnt=config.max_vocab_cnt,
with_sentiment=with_sentiment)
# 交替训练，准备大数据集
poem_corpus = api.get_tokenized_poem_corpus(
type=1 + int(with_sentiment)) # corpus for training and validation
test_data = api.get_tokenized_test_corpus() # 测试数据
# 三个list，每个list中的每一个元素都是 [topic, last_sentence, current_sentence]
train_poem, valid_poem, test_poem = poem_corpus["train"], poem_corpus[
"valid"], test_data["test"]
train_loader = SWDADataLoader("Train", train_poem, config)
valid_loader = SWDADataLoader("Valid", valid_poem, config)
test_loader = SWDADataLoader("Test", test_poem, config)
print("Finish Poem data loading, not pretraining or alignment test")
if not args.forward_only:
# LOG #
log_start_time = str(datetime.now().strftime('%Y%m%d%H%M'))
if not os.path.isdir('./output'):
os.makedirs('./output')
if not os.path.isdir('./output/{}'.format(args.expname)):
os.makedirs('./output/{}'.format(args.expname))
if not os.path.isdir('./output/{}/{}'.format(args.expname,
log_start_time)):
os.makedirs('./output/{}/{}'.format(args.expname, log_start_time))
# save arguments
json.dump(
vars(args),
open(
'./output/{}/{}/args.json'.format(args.expname,
log_start_time), 'w'))
logger = logging.getLogger(__name__)
logging.basicConfig(level=logging.DEBUG, format="%(message)s")
fh = logging.FileHandler("./output/{}/{}/logs.txt".format(
args.expname, log_start_time))
# add the handlers to the logger
logger.addHandler(fh)
logger.info(vars(args))
tb_writer = SummaryWriter("./output/{}/{}/tb_logs".format(
args.expname, log_start_time)) if args.visual else None
if config.reload_model:
model = load_model(config.model_name)
else:
if args.model == "mCVAE":
model = CVAE_GMP(config=config, api=api)
elif args.model == 'CVAE':
model = CVAE(config=config, api=api)
else:
model = Seq2Seq(config=config, api=api)
if use_cuda:
model = model.cuda()
# if corpus.word2vec is not None and args.reload_from<0:
# print("Loaded word2vec")
# model.embedder.weight.data.copy_(torch.from_numpy(corpus.word2vec))
# model.embedder.weight.data[0].fill_(0)
###############################################################################
# Start training
###############################################################################
# model依然是PoemWAE_GMP保持不变，只不过，用这部分数据强制训练其中一个高斯先验分布
# pretrain = True
cur_best_score = {
'min_valid_loss': 100,
'min_global_itr': 0,
'min_epoch': 0,
'min_itr': 0
}
train_loader.epoch_init(config.batch_size, shuffle=True)
# model = load_model(3, 3)
epoch_id = 0
global_t = 0
while epoch_id < config.epochs:
while True: # loop through all batches in training data
# train一个batch
model, finish_train, loss_records, global_t = \
train_process(global_t=global_t, model=model, train_loader=train_loader, config=config, sentiment_data=with_sentiment)
if finish_train:
test_process(model=model,
test_loader=test_loader,
test_config=test_config,
logger=logger)
# evaluate_process(model=model, valid_loader=valid_loader, log_start_time=log_start_time, global_t=global_t, epoch=epoch_id, logger=logger, tb_writer=tb_writer, api=api)
# save model after each epoch
save_model(model=model,
epoch=epoch_id,
global_t=global_t,
log_start_time=log_start_time)
logger.info(
'Finish epoch %d, current min valid loss: %.4f \
correspond epoch: %d itr: %d \n\n' %
(cur_best_score['min_valid_loss'],
cur_best_score['min_global_itr'],
cur_best_score['min_epoch'],
cur_best_score['min_itr']))
# 初始化下一个unlabeled data epoch的训练
# unlabeled_epoch += 1
epoch_id += 1
train_loader.epoch_init(config.batch_size, shuffle=True)
break
# elif batch_idx >= start_batch + config.n_batch_every_iter:
# print("Finish unlabel epoch %d batch %d to %d" %
# (unlabeled_epoch, start_batch, start_batch + config.n_batch_every_iter))
# start_batch += config.n_batch_every_iter
# break
# 写一下log
if global_t % config.log_every == 0:
log = 'Epoch id %d: step: %d/%d: ' \
% (epoch_id, global_t % train_loader.num_batch, train_loader.num_batch)
for loss_name, loss_value in loss_records:
if loss_name == 'avg_lead_loss':
continue
log = log + loss_name + ':%.4f ' % loss_value
if args.visual:
tb_writer.add_scalar(loss_name, loss_value,
global_t)
logger.info(log)
# valid
if global_t % config.valid_every == 0:
# test_process(model=model, test_loader=test_loader, test_config=test_config, logger=logger)
valid_process(
global_t=global_t,
model=model,
valid_loader=valid_loader,
valid_config=valid_config,
unlabeled_epoch=
epoch_id, # 如果sample_rate_unlabeled不是1，这里要在最后加一个1
tb_writer=tb_writer,
logger=logger,
cur_best_score=cur_best_score)
# if batch_idx % (train_loader.num_batch // 3) == 0:
# test_process(model=model, test_loader=test_loader, test_config=test_config, logger=logger)
if global_t % config.test_every == 0:
test_process(model=model,
test_loader=test_loader,
test_config=test_config,
logger=logger)
# forward_only 测试
else:
expname = 'sentInput'
time = '202101191105'
model = load_model(
'./output/{}/{}/model_global_t_13596_epoch3.pckl'.format(
expname, time))
test_loader.epoch_init(test_config.batch_size, shuffle=False)
if not os.path.exists('./output/{}/{}/test/'.format(expname, time)):
os.mkdir('./output/{}/{}/test/'.format(expname, time))
output_file = [
open('./output/{}/{}/test/output_0.txt'.format(expname, time),
'w'),
open('./output/{}/{}/test/output_1.txt'.format(expname, time),
'w'),
open('./output/{}/{}/test/output_2.txt'.format(expname, time), 'w')
]
poem_count = 0
predict_results = {0: [], 1: [], 2: []}
titles = {0: [], 1: [], 2: []}
sentiment_result = {0: [], 1: [], 2: []}
# Get all poem predictions
while True:
model.eval()
batch = test_loader.next_batch_test() # test data使用专门的batch
poem_count += 1
if poem_count % 10 == 0:
print("Predicted {} poems".format(poem_count))
if batch is None:
break
title_list = batch # batch size是1，一个batch写一首诗
title_tensor = to_tensor(title_list)
# test函数将当前batch对应的这首诗decode出来，记住每次decode的输入context是上一次的结果
for i in range(3):
sentiment_label = np.zeros(1, dtype=np.int64)
sentiment_label[0] = int(i)
sentiment_label = to_tensor(sentiment_label)
output_poem, output_tokens = model.test(
title_tensor, title_list, sentiment_label=sentiment_label)
titles[i].append(output_poem.strip().split('\n')[0])
predict_results[i] += (np.array(output_tokens)[:, :7].tolist())
# Predict sentiment use the sort net
from collections import defaultdict
neg = defaultdict(int)
neu = defaultdict(int)
pos = defaultdict(int)
total = defaultdict(int)
for i in range(3):
_, neg[i], neu[i], pos[i] = test_sentiment(predict_results[i])
total[i] = neg[i] + neu[i] + pos[i]
for i in range(3):
print("%d%%\t%d%%\t%d%%" %
(neg * 100 / total, neu * 100 / total, pos * 100 / total))
for i in range(3):
write_predict_result_to_file(titles[i], predict_results[i],
sentiment_result[i], output_file[i])
output_file[i].close()
print("Done testing")
def main():
# config for training
config = Config()
# config for validation
valid_config = Config()
valid_config.keep_prob = 1.0
valid_config.dec_keep_prob = 1.0
valid_config.batch_size = 60
# configuration for testing
test_config = Config()
test_config.keep_prob = 1.0
test_config.dec_keep_prob = 1.0
test_config.batch_size = 1
pp(config)
# get data set
api = SWDADialogCorpus(FLAGS.data_dir,
word2vec=FLAGS.word2vec_path,
word2vec_dim=config.embed_size)
dial_corpus = api.get_dialog_corpus()
meta_corpus = api.get_meta_corpus()
train_meta, valid_meta, test_meta = meta_corpus.get(
"train"), meta_corpus.get("valid"), meta_corpus.get("test")
train_dial, valid_dial, test_dial = dial_corpus.get(
"train"), dial_corpus.get("valid"), dial_corpus.get("test")
# convert to numeric input outputs that fits into TF models
train_feed = SWDADataLoader("Train", train_dial, train_meta, config)
valid_feed = SWDADataLoader("Valid", valid_dial, valid_meta, config)
test_feed = SWDADataLoader("Test", test_dial, test_meta, config)
if FLAGS.forward_only or FLAGS.resume:
log_dir = os.path.join(FLAGS.work_dir, FLAGS.test_path)
else:
log_dir = os.path.join(FLAGS.work_dir, "run" + str(int(time.time())))
# begin training
if True:
scope = "model"
model = KgRnnCVAE(config,
api,
log_dir=None if FLAGS.forward_only else log_dir,
scope=scope)
print("Created computation graphs")
# write config to a file for logging
if not FLAGS.forward_only:
with open(os.path.join(log_dir, "run.log"), "wb") as f:
f.write(pp(config, output=False))
# create a folder by force
ckp_dir = os.path.join(log_dir, "checkpoints")
if not os.path.exists(ckp_dir):
os.mkdir(ckp_dir)
ckpt = get_checkpoint_state(ckp_dir)
print("Created models with fresh parameters.")
model.apply(lambda m: [
torch.nn.init.uniform(p.data, -1.0 * config.init_w, config.init_w)
for p in m.parameters()
])
# Load word2vec weight
if api.word2vec is not None and not FLAGS.forward_only:
print("Loaded word2vec")
model.embedding.weight.data.copy_(
torch.from_numpy(np.array(api.word2vec)))
model.embedding.weight.data[0].fill_(0)
if ckpt:
print("Reading dm models parameters from %s" % ckpt)
model.load_state_dict(torch.load(ckpt))
#
# turn to cuda
model.cuda()
if not FLAGS.forward_only:
dm_checkpoint_path = os.path.join(
ckp_dir, model.__class__.__name__ + "-%d.pth")
global_t = 1
patience = 10 # wait for at least 10 epoch before stop
dev_loss_threshold = np.inf
best_dev_loss = np.inf
for epoch in range(config.max_epoch):
print(">> Epoch %d with lr %f" % (epoch, model.learning_rate))
# begin training
if train_feed.num_batch is None or train_feed.ptr >= train_feed.num_batch:
train_feed.epoch_init(config.batch_size,
config.backward_size,
config.step_size,
shuffle=True)
global_t, train_loss = model.train_model(
global_t, train_feed, update_limit=config.update_limit)
# begin validation
valid_feed.epoch_init(valid_config.batch_size,
valid_config.backward_size,
valid_config.step_size,
shuffle=False,
intra_shuffle=False)
model.eval()
valid_loss = model.valid_model("ELBO_VALID", valid_feed)
test_feed.epoch_init(test_config.batch_size,
test_config.backward_size,
test_config.step_size,
shuffle=True,
intra_shuffle=False)
model.test_model(test_feed, num_batch=5)
model.train()
done_epoch = epoch + 1
# only save a models if the dev loss is smaller
# Decrease learning rate if no improvement was seen over last 3 times.
if config.op == "sgd" and done_epoch > config.lr_hold:
model.learning_rate_decay()
if valid_loss < best_dev_loss:
if valid_loss <= dev_loss_threshold * config.improve_threshold:
patience = max(patience,
done_epoch * config.patient_increase)
dev_loss_threshold = valid_loss
# still save the best train model
if FLAGS.save_model:
print("Save model!!")
torch.save(model.state_dict(),
dm_checkpoint_path % (epoch))
best_dev_loss = valid_loss
if config.early_stop and patience <= done_epoch:
print("!!Early stop due to run out of patience!!")
break
print("Best validation loss %f" % best_dev_loss)
print("Done training")
else:
# begin validation
# begin validation
valid_feed.epoch_init(valid_config.batch_size,
valid_config.backward_size,
valid_config.step_size,
shuffle=False,
intra_shuffle=False)
model.eval()
model.valid_model("ELBO_VALID", valid_feed)
test_feed.epoch_init(valid_config.batch_size,
valid_config.backward_size,
valid_config.step_size,
shuffle=False,
intra_shuffle=False)
model.valid_model("ELBO_TEST", test_feed)
dest_f = open(os.path.join(log_dir, "test.txt"), "wb")
test_feed.epoch_init(test_config.batch_size,
test_config.backward_size,
test_config.step_size,
shuffle=False,
intra_shuffle=False)
model.test_model(test_feed, num_batch=None, repeat=10, dest=dest_f)
model.train()
dest_f.close()
def main():
# config for training
config = Config()
# config for validation
valid_config = Config()
valid_config.keep_prob = 1.0
valid_config.dec_keep_prob = 1.0
valid_config.batch_size = 60
# configuration for testing
test_config = Config()
test_config.keep_prob = 1.0
test_config.dec_keep_prob = 1.0
test_config.batch_size = 1
pp(config)
# get data set
api = SWDADialogCorpus(FLAGS.data_dir, word2vec=FLAGS.word2vec_path, word2vec_dim=config.embed_size)
dial_corpus = api.get_dialog_corpus()
meta_corpus = api.get_meta_corpus()
'''item in meta like:
([0.46, 0.6666666666666666, 1, 0], [0.48, 0.6666666666666666, 1, 0], 29)
'''
train_meta, valid_meta, test_meta = meta_corpus.get("train"), meta_corpus.get("valid"), meta_corpus.get("test")
'''item in dial
([utt_id,], spearker_id, [dialogact_id, [,,,]])
'''
train_dial, valid_dial, test_dial = dial_corpus.get("train"), dial_corpus.get("valid"), dial_corpus.get("test")
# convert to numeric input outputs that fits into TF models
train_feed = SWDADataLoader("Train", train_dial, train_meta, config)
valid_feed = SWDADataLoader("Valid", valid_dial, valid_meta, config)
test_feed = SWDADataLoader("Test", test_dial, test_meta, config)
if FLAGS.forward_only or FLAGS.resume:
log_dir = os.path.join(FLAGS.work_dir, FLAGS.test_path)
else:
log_dir = os.path.join(FLAGS.work_dir, "run"+str(int(time.time())))
# begin training
with tf.Session() as sess:
initializer = tf.random_uniform_initializer(-1.0 * config.init_w, config.init_w)
scope = "model"
with tf.variable_scope(scope, reuse=None, initializer=initializer):
model = KgRnnCVAE(sess, config, api, log_dir=None if FLAGS.forward_only else log_dir, forward=False, scope=scope)
with tf.variable_scope(scope, reuse=True, initializer=initializer):
valid_model = KgRnnCVAE(sess, valid_config, api, log_dir=None, forward=False, scope=scope)
with tf.variable_scope(scope, reuse=True, initializer=initializer):
test_model = KgRnnCVAE(sess, test_config, api, log_dir=None, forward=True, scope=scope)
print("Created computation graphs")
if api.word2vec is not None and not FLAGS.forward_only:
print("Loaded word2vec")
sess.run(model.embedding.assign(np.array(api.word2vec)))
# write config to a file for logging
if not FLAGS.forward_only:
with open(os.path.join(log_dir, "run.log"), "wb") as f:
f.write(pp(config, output=False))
# create a folder by force
ckp_dir = os.path.join(log_dir, "checkpoints")
if not os.path.exists(ckp_dir):
os.mkdir(ckp_dir)
ckpt = tf.train.get_checkpoint_state(ckp_dir)
print("Created models with fresh parameters.")
sess.run(tf.global_variables_initializer())
if ckpt:
print("Reading dm models parameters from %s" % ckpt.model_checkpoint_path)
model.saver.restore(sess, ckpt.model_checkpoint_path)
if not FLAGS.forward_only:
dm_checkpoint_path = os.path.join(ckp_dir, model.__class__.__name__+ ".ckpt")
global_t = 1
patience = 10 # wait for at least 10 epoch before stop
dev_loss_threshold = np.inf
best_dev_loss = np.inf
for epoch in range(config.max_epoch):
print(">> Epoch %d with lr %f" % (epoch, model.learning_rate.eval()))
# begin training
if train_feed.num_batch is None or train_feed.ptr >= train_feed.num_batch:
train_feed.epoch_init(config.batch_size, config.backward_size,
config.step_size, shuffle=True)
global_t, train_loss = model.train(global_t, sess, train_feed, update_limit=config.update_limit)
# begin validation
valid_feed.epoch_init(valid_config.batch_size, valid_config.backward_size,
valid_config.step_size, shuffle=False, intra_shuffle=False)
valid_loss = valid_model.valid("ELBO_VALID", sess, valid_feed)
test_feed.epoch_init(test_config.batch_size, test_config.backward_size,
test_config.step_size, shuffle=True, intra_shuffle=False)
test_model.test(sess, test_feed, num_batch=5)
done_epoch = epoch + 1
# only save a models if the dev loss is smaller
# Decrease learning rate if no improvement was seen over last 3 times.
if config.op == "sgd" and done_epoch > config.lr_hold:
sess.run(model.learning_rate_decay_op)
if valid_loss < best_dev_loss:
if valid_loss <= dev_loss_threshold * config.improve_threshold:
patience = max(patience, done_epoch * config.patient_increase)
dev_loss_threshold = valid_loss
# still save the best train model
if FLAGS.save_model:
print("Save model!!")
model.saver.save(sess, dm_checkpoint_path, global_step=epoch)
best_dev_loss = valid_loss
if config.early_stop and patience <= done_epoch:
print("!!Early stop due to run out of patience!!")
break
print("Best validation loss %f" % best_dev_loss)
print("Done training")
elif not FLAGS.demo:
# begin validation
# begin validation
valid_feed.epoch_init(valid_config.batch_size, valid_config.backward_size,
valid_config.step_size, shuffle=False, intra_shuffle=False)
valid_model.valid("ELBO_VALID", sess, valid_feed)
test_feed.epoch_init(valid_config.batch_size, valid_config.backward_size,
valid_config.step_size, shuffle=False, intra_shuffle=False)
valid_model.valid("ELBO_TEST", sess, test_feed)
dest_f = open(os.path.join(log_dir, "test.txt"), "wb")
test_feed.epoch_init(test_config.batch_size, test_config.backward_size,
test_config.step_size, shuffle=False, intra_shuffle=False)
test_model.test(sess, test_feed, num_batch=None, repeat=10, dest=dest_f)
dest_f.close()
else:
infer_api = Corpus(FLAGS.vocab, FLAGS.da_vocab, FLAGS.topic_vocab)
while(1):
demo_sent = raw_input('Please input your sentence:\t')
if demo_sent == '' or demo_sent.isspace():
print('See you next time!')
break
speaker = raw_input('Are you speaker A?: 1 or 0\t')
topic = raw_input('what is your topic:\t')
# demo_sent = 'Hello'
# speaker = '1'
# topic = 'MUSIC'
meta, dial = infer_api.format_input(demo_sent, speaker, topic)
infer_feed = SWDADataLoader("Infer", dial, meta, config)
infer_feed.epoch_init(test_config.batch_size, test_config.backward_size,
test_config.step_size, shuffle=False, intra_shuffle=False)
pred_strs, pred_das = test_model.inference(sess, infer_feed, num_batch=None, repeat=3)
if pred_strs and pred_das:
print('Here is your answer')
for da, answer in zip(pred_strs, pred_das):
print('{} >> {}'.format(da, answer))
def main():
## random seeds
seed = FLAGS.seed
# tf.random.set_seed(seed)
np.random.seed(seed)
## config for training
config = Config()
pid = PIDControl(FLAGS.exp_KL)
# config for validation
valid_config = Config()
valid_config.keep_prob = 1.0
valid_config.dec_keep_prob = 1.0
valid_config.batch_size = 60
# configuration for testing
test_config = Config()
test_config.keep_prob = 1.0
test_config.dec_keep_prob = 1.0
test_config.batch_size = 1
pp(config)
# get data set
api = SWDADialogCorpus(FLAGS.data_dir, word2vec=FLAGS.word2vec_path, word2vec_dim=config.embed_size)
dial_corpus = api.get_dialog_corpus()
meta_corpus = api.get_meta_corpus()
train_meta, valid_meta, test_meta = meta_corpus.get("train"), meta_corpus.get("valid"), meta_corpus.get("test")
train_dial, valid_dial, test_dial = dial_corpus.get("train"), dial_corpus.get("valid"), dial_corpus.get("test")
# convert to numeric input outputs that fits into TF models
train_feed = SWDADataLoader("Train", train_dial, train_meta, config)
valid_feed = SWDADataLoader("Valid", valid_dial, valid_meta, config)
test_feed = SWDADataLoader("Test", test_dial, test_meta, config)
if FLAGS.forward_only or FLAGS.resume:
# log_dir = os.path.join(FLAGS.work_dir, FLAGS.test_path)
log_dir = os.path.join(FLAGS.work_dir, FLAGS.model_name)
else:
log_dir = os.path.join(FLAGS.work_dir, FLAGS.model_name)
## begin training
with tf.Session() as sess:
initializer = tf.random_uniform_initializer(-1.0 * config.init_w, config.init_w)
scope = "model"
with tf.variable_scope(scope, reuse=None, initializer=initializer):
model = KgRnnCVAE(sess, config, api, log_dir=None if FLAGS.forward_only else log_dir, forward=False, pid_control=pid, scope=scope)
with tf.variable_scope(scope, reuse=True, initializer=initializer):
valid_model = KgRnnCVAE(sess, valid_config, api, log_dir=None, forward=False, pid_control=pid, scope=scope)
with tf.variable_scope(scope, reuse=True, initializer=initializer):
test_model = KgRnnCVAE(sess, test_config, api, log_dir=None, forward=True, pid_control=pid, scope=scope)
print("Created computation graphs")
if api.word2vec is not None and not FLAGS.forward_only:
print("Loaded word2vec")
sess.run(model.embedding.assign(np.array(api.word2vec)))
# write config to a file for logging
if not FLAGS.forward_only:
with open(os.path.join(log_dir, "configure.log"), "wb") as f:
f.write(pp(config, output=False))
# create a folder by force
ckp_dir = os.path.join(log_dir, "checkpoints")
print("*******checkpoint path: ", ckp_dir)
if not os.path.exists(ckp_dir):
os.mkdir(ckp_dir)
ckpt = tf.train.get_checkpoint_state(ckp_dir)
print("Created models with fresh parameters.")
sess.run(tf.global_variables_initializer())
if ckpt:
print("Reading dm models parameters from %s" % ckpt.model_checkpoint_path)
model.saver.restore(sess, ckpt.model_checkpoint_path)
### save log when running
if not FLAGS.forward_only:
logfileName = "train.log"
else:
logfileName = "test.log"
fw_log = open(os.path.join(log_dir, logfileName), "w")
print("log directory >>> : ", os.path.join(log_dir, "run.log"))
if not FLAGS.forward_only:
print('--start training now---')
dm_checkpoint_path = os.path.join(ckp_dir, model.__class__.__name__+ ".ckpt")
global_t = 1
patience = 20 # wait for at least 10 epoch before stop
dev_loss_threshold = np.inf
best_dev_loss = np.inf
pbar = tqdm(total = config.max_epoch)
## epoch start training
for epoch in range(config.max_epoch):
pbar.update(1)
print(">> Epoch %d with lr %f" % (epoch, model.learning_rate.eval()))
## begin training
FLAGS.mode = 'train'
if train_feed.num_batch is None or train_feed.ptr >= train_feed.num_batch:
train_feed.epoch_init(config.batch_size, config.backward_size,
config.step_size, shuffle=True)
global_t, train_loss = model.train(global_t, sess, train_feed, update_limit=config.update_limit)
FLAGS.mode = 'valid'
valid_feed.epoch_init(valid_config.batch_size, valid_config.backward_size,
valid_config.step_size, shuffle=False, intra_shuffle=False)
test_feed.epoch_init(valid_config.batch_size, valid_config.backward_size,
valid_config.step_size, shuffle=False, intra_shuffle=False)
elbo, nll, ppl, au_count, kl_loss = valid_model.valid("ELBO_TEST", sess, valid_feed, test_feed)
print('middle test nll: {} ppl: {} ActiveUnit: {} kl_loss:{}\n'.format(nll, ppl,au_count,kl_loss))
fw_log.write('epoch:{} testing nll:{} ppl:{} ActiveUnit:{} kl_loss:{} elbo:{}\n'.\
format(epoch, nll, ppl, au_count, kl_loss, elbo))
fw_log.flush()
'''
## begin validation
FLAGS.mode = 'valid'
valid_feed.epoch_init(valid_config.batch_size, valid_config.backward_size,
valid_config.step_size, shuffle=False, intra_shuffle=False)
valid_loss = valid_model.valid("ELBO_VALID", sess, valid_feed)
## test model
FLAGS.mode = 'test'
test_feed.epoch_init(test_config.batch_size, test_config.backward_size,
test_config.step_size, shuffle=True, intra_shuffle=False)
test_model.test(sess, test_feed, num_batch=5)
done_epoch = epoch + 1
# only save a models if the dev loss is smaller
# Decrease learning rate if no improvement was seen over last 3 times.
if config.op == "sgd" and done_epoch > config.lr_hold:
sess.run(model.learning_rate_decay_op)
if valid_loss < best_dev_loss:
if valid_loss <= dev_loss_threshold * config.improve_threshold:
patience = max(patience, done_epoch * config.patient_increase)
dev_loss_threshold = valid_loss
# still save the best train model
if FLAGS.save_model:
print("Save model!!")
model.saver.save(sess, dm_checkpoint_path, global_step=epoch)
best_dev_loss = valid_loss
if config.early_stop and patience <= done_epoch:
print("!!Early stop due to run out of patience!!")
break
## print("Best validation loss %f" % best_dev_loss)
'''
print("Done training and save checkpoint")
if FLAGS.save_model:
print("Save model!!")
model.saver.save(sess, dm_checkpoint_path, global_step=epoch)
# begin validation
print('--------after training to testing now-----')
FLAGS.mode = 'test'
# valid_feed.epoch_init(valid_config.batch_size, valid_config.backward_size,
# valid_config.step_size, shuffle=False, intra_shuffle=False)
# valid_model.valid("ELBO_VALID", sess, valid_feed)
valid_feed.epoch_init(valid_config.batch_size, valid_config.backward_size,
valid_config.step_size, shuffle=False, intra_shuffle=False)
test_feed.epoch_init(valid_config.batch_size, valid_config.backward_size,
valid_config.step_size, shuffle=False, intra_shuffle=False)
elbo, nll, ppl, au_count,kl_loss = valid_model.valid("ELBO_TEST", sess, valid_feed, test_feed)
print('final test nll: {} ppl: {} ActiveUnit: {} kl_loss:{}\n'.format(nll, ppl,au_count,kl_loss))
fw_log.write('Final testing nll:{} ppl:{} ActiveUnit:{} kl_loss:{} elbo:{}\n'.\
format(nll, ppl, au_count, kl_loss, elbo))
dest_f = open(os.path.join(log_dir, FLAGS.test_res), "wb")
test_feed.epoch_init(test_config.batch_size, test_config.backward_size,
test_config.step_size, shuffle=False, intra_shuffle=False)
test_model.test(sess, test_feed, num_batch=None, repeat=10, dest=dest_f)
dest_f.close()
print("****testing done****")
else:
# begin validation
# begin validation
print('*'*89)
print('--------testing now-----')
print('*'*89)
FLAGS.mode = 'test'
valid_feed.epoch_init(valid_config.batch_size, valid_config.backward_size,
valid_config.step_size, shuffle=False, intra_shuffle=False)
# valid_model.valid("ELBO_VALID", sess, valid_feed)
test_feed.epoch_init(valid_config.batch_size, valid_config.backward_size,
valid_config.step_size, shuffle=False, intra_shuffle=False)
elbo, nll, ppl, au_count, kl_loss = valid_model.valid("ELBO_TEST", sess, valid_feed, test_feed)
print('final test nll: {} ppl: {} ActiveUnit: {} kl_loss:{}\n'.format(nll, ppl,au_count,kl_loss))
fw_log.write('Final testing nll:{} ppl:{} ActiveUnit:{} kl_loss:{} elbo:{}\n'.\
format(nll, ppl, au_count, kl_loss, elbo))
# dest_f = open(os.path.join(log_dir, FLAGS.test_res), "wb")
# test_feed.epoch_init(test_config.batch_size, test_config.backward_size,
# test_config.step_size, shuffle=False, intra_shuffle=False)
# test_model.test(sess, test_feed, num_batch=None, repeat=10, dest=dest_f)
# dest_f.close()
print("****testing done****")
fw_log.close()
def main():
# config for training
config = Config()
print("Normal train config:")
pp(config)
valid_config = Config()
valid_config.dropout = 0
valid_config.batch_size = 20
# config for test
test_config = Config()
test_config.dropout = 0
test_config.batch_size = 1
with_sentiment = config.with_sentiment
pretrain = False
###############################################################################
# Logs
###############################################################################
log_start_time = str(datetime.now().strftime('%Y%m%d%H%M'))
if not os.path.isdir('./output'):
os.makedirs('./output')
if not os.path.isdir('./output/{}'.format(args.expname)):
os.makedirs('./output/{}'.format(args.expname))
if not os.path.isdir('./output/{}/{}'.format(args.expname,
log_start_time)):
os.makedirs('./output/{}/{}'.format(args.expname, log_start_time))
# save arguments
json.dump(
vars(args),
open('./output/{}/{}/args.json'.format(args.expname, log_start_time),
'w'))
logger = logging.getLogger(__name__)
logging.basicConfig(level=logging.DEBUG, format="%(message)s")
fh = logging.FileHandler("./output/{}/{}/logs.txt".format(
args.expname, log_start_time))
# add the handlers to the logger
logger.addHandler(fh)
logger.info(vars(args))
tb_writer = SummaryWriter("./output/{}/{}/tb_logs".format(
args.expname, log_start_time)) if args.visual else None
###############################################################################
# Model
###############################################################################
# vocab and rev_vocab
with open(args.vocab_path) as vocab_file:
vocab = vocab_file.read().strip().split('\n')
rev_vocab = {vocab[idx]: idx for idx in range(len(vocab))}
if not pretrain:
pass
# assert config.reload_model
# model = load_model(config.model_name)
else:
if args.model == "multiVAE":
model = multiVAE(config=config, vocab=vocab, rev_vocab=rev_vocab)
else:
model = CVAE(config=config, vocab=vocab, rev_vocab=rev_vocab)
if use_cuda:
model = model.cuda()
###############################################################################
# Load data
###############################################################################
if pretrain:
from collections import defaultdict
api = LoadPretrainPoem(corpus_path=args.pretrain_data_dir,
vocab_path="data/vocab.txt")
train_corpus, valid_corpus = defaultdict(list), defaultdict(list)
divide = 50000
train_corpus['pos'], valid_corpus['pos'] = api.data[
'pos'][:divide], api.data['pos'][divide:]
train_corpus['neu'], valid_corpus['neu'] = api.data[
'neu'][:divide], api.data['neu'][divide:]
train_corpus['neg'], valid_corpus['neg'] = api.data[
'neg'][:divide], api.data['neg'][divide:]
token_corpus = defaultdict(dict)
token_corpus['pos'], token_corpus['neu'], token_corpus['neg'] = \
api.get_tokenized_poem_corpus(train_corpus['pos'], valid_corpus['pos']), \
api.get_tokenized_poem_corpus(train_corpus['neu'], valid_corpus['neu']), \
api.get_tokenized_poem_corpus(train_corpus['neg'], valid_corpus['neg']),
# train_loader_dict = {'pos': }
train_loader = {
'pos': SWDADataLoader("Train", token_corpus['pos']['train'],
config),
'neu': SWDADataLoader("Train", token_corpus['neu']['train'],
config),
'neg': SWDADataLoader("Train", token_corpus['neg']['train'],
config)
}
valid_loader = {
'pos': SWDADataLoader("Train", token_corpus['pos']['valid'],
config),
'neu': SWDADataLoader("Train", token_corpus['neu']['valid'],
config),
'neg': SWDADataLoader("Train", token_corpus['neg']['valid'],
config)
}
###############################################################################
# Pretrain three VAEs
###############################################################################
epoch_id = 0
global_t = 0
init_train_loaders(train_loader, config)
while epoch_id < config.epochs:
while True: # loop through all batches in training data
# train一个batch
model, finish_train, loss_records, global_t = \
pre_train_process(global_t=global_t, model=model, train_loader=train_loader)
if finish_train:
if epoch_id > 5:
save_model(model=model,
epoch=epoch_id,
global_t=global_t,
log_start_time=log_start_time)
epoch_id += 1
init_train_loaders(train_loader, config)
break
# 写一下log
if global_t % config.log_every == 0:
pre_log_process(epoch_id=epoch_id,
global_t=global_t,
train_loader=train_loader,
loss_records=loss_records,
logger=logger,
tb_writer=tb_writer)
# valid
if global_t % config.valid_every == 0:
# test_process(model=model, test_loader=test_loader, test_config=test_config, logger=logger)
pre_valid_process(global_t=global_t,
model=model,
valid_loader=valid_loader,
valid_config=valid_config,
tb_writer=tb_writer,
logger=logger)
if global_t % config.test_every == 0:
pre_test_process(model=model, logger=logger)
###############################################################################
# Train the big model
###############################################################################
api = LoadPoem(corpus_path=args.train_data_dir,
vocab_path="data/vocab.txt",
test_path=args.test_data_dir,
max_vocab_cnt=config.max_vocab_cnt,
with_sentiment=with_sentiment)
from collections import defaultdict
token_corpus = defaultdict(dict)
token_corpus['pos'], token_corpus['neu'], token_corpus['neg'] = \
api.get_tokenized_poem_corpus(api.train_corpus['pos'], api.valid_corpus['pos']), \
api.get_tokenized_poem_corpus(api.train_corpus['neu'], api.valid_corpus['neu']), \
api.get_tokenized_poem_corpus(api.train_corpus['neg'], api.valid_corpus['neg']),
train_loader = {
'pos': SWDADataLoader("Train", token_corpus['pos']['train'], config),
'neu': SWDADataLoader("Train", token_corpus['neu']['train'], config),
'neg': SWDADataLoader("Train", token_corpus['neg']['train'], config)
}
valid_loader = {
'pos': SWDADataLoader("Train", token_corpus['pos']['valid'], config),
'neu': SWDADataLoader("Train", token_corpus['neu']['valid'], config),
'neg': SWDADataLoader("Train", token_corpus['neg']['valid'], config)
}
test_poem = api.get_tokenized_test_corpus()['test'] # 测试数据
test_loader = SWDADataLoader("Test", test_poem, config)
print("Finish Poem data loading, not pretraining or alignment test")
if not args.forward_only:
# model依然是PoemWAE_GMP保持不变，只不过，用这部分数据强制训练其中一个高斯先验分布
# pretrain = True
cur_best_score = {
'min_valid_loss': 100,
'min_global_itr': 0,
'min_epoch': 0,
'min_itr': 0
}
# model = load_model(3, 3)
epoch_id = 0
global_t = 0
init_train_loaders(train_loader, config)
while epoch_id < config.epochs:
while True: # loop through all batches in training data
# train一个batch
model, finish_train, loss_records, global_t = \
train_process(global_t=global_t, model=model, train_loader=train_loader)
if finish_train:
if epoch_id > 5:
save_model(model=model,
epoch=epoch_id,
global_t=global_t,
log_start_time=log_start_time)
epoch_id += 1
init_train_loaders(train_loader, config)
break
# 写一下log
if global_t % config.log_every == 0:
pre_log_process(epoch_id=epoch_id,
global_t=global_t,
train_loader=train_loader,
loss_records=loss_records,
logger=logger,
tb_writer=tb_writer)
# valid
if global_t % config.valid_every == 0:
valid_process(global_t=global_t,
model=model,
valid_loader=valid_loader,
valid_config=valid_config,
tb_writer=tb_writer,
logger=logger)
# if batch_idx % (train_loader.num_batch // 3) == 0:
# test_process(model=model, test_loader=test_loader, test_config=test_config, logger=logger)
if global_t % config.test_every == 0:
test_process(model=model,
test_loader=test_loader,
test_config=test_config,
logger=logger)
# forward_only 测试
else:
expname = 'trainVAE'
time = '202101231631'
model = load_model(
'./output/{}/{}/model_global_t_26250_epoch9.pckl'.format(
expname, time))
test_loader.epoch_init(test_config.batch_size, shuffle=False)
if not os.path.exists('./output/{}/{}/test/'.format(expname, time)):
os.mkdir('./output/{}/{}/test/'.format(expname, time))
output_file = [
open('./output/{}/{}/test/output_0.txt'.format(expname, time),
'w'),
open('./output/{}/{}/test/output_1.txt'.format(expname, time),
'w'),
open('./output/{}/{}/test/output_2.txt'.format(expname, time), 'w')
]
poem_count = 0
predict_results = {0: [], 1: [], 2: []}
titles = {0: [], 1: [], 2: []}
sentiment_result = {0: [], 1: [], 2: []}
# sent_dict = {0: ['0', '1', '1', '0'], 1: ['2', '1', '2', '2'], 2: ['1', '0', '1', '2']}
sent_dict = {
0: ['0', '0', '0', '0'],
1: ['1', '1', '1', '1'],
2: ['2', '2', '2', '2']
}
# Get all poem predictions
while True:
model.eval()
batch = test_loader.next_batch_test() # test data使用专门的batch
poem_count += 1
if poem_count % 10 == 0:
print("Predicted {} poems".format(poem_count))
if batch is None:
break
title_list = batch # batch size是1，一个batch写一首诗
title_tensor = to_tensor(title_list)
# test函数将当前batch对应的这首诗decode出来，记住每次decode的输入context是上一次的结果
for i in range(3):
sent_labels = sent_dict[i]
for _ in range(4):
sent_labels.append(str(i))
output_poem, output_tokens = model.test(
title_tensor, title_list, sent_labels=sent_labels)
titles[i].append(output_poem.strip().split('\n')[0])
predict_results[i] += (np.array(output_tokens)[:, :7].tolist())
# Predict sentiment use the sort net
from collections import defaultdict
neg = defaultdict(int)
neu = defaultdict(int)
pos = defaultdict(int)
total = defaultdict(int)
for i in range(3):
cur_sent_result, neg[i], neu[i], pos[i] = test_sentiment(
predict_results[i])
sentiment_result[i] = cur_sent_result
total[i] = neg[i] + neu[i] + pos[i]
for i in range(3):
print("%d%%\t%d%%\t%d%%" % (neg[i] * 100 / total[i], neu[i] * 100 /
total[i], pos[i] * 100 / total[i]))
for i in range(3):
write_predict_result_to_file(titles[i], predict_results[i],
sentiment_result[i], output_file[i])
output_file[i].close()
print("Done testing")
def main(model_type):
# config for training
config = Config()
# config for validation
valid_config = Config()
valid_config.keep_prob = 1.0
valid_config.dec_keep_prob = 1.0
valid_config.batch_size = 60
# configuration for testing
test_config = Config()
test_config.keep_prob = 1.0
test_config.dec_keep_prob = 1.0
test_config.batch_size = 1
pp(config)
# which model to run
if model_type == "kgcvae":
model_class = KgRnnCVAE
backward_size = config.backward_size
config.use_hcf, valid_config.use_hcf, test_config.use_hcf = True
elif model_type == "cvae":
model_class = KgRnnCVAE
backward_size = config.backward_size
config.use_hcf, valid_config.use_hcf, test_config.use_hcf = False
elif model_type == 'hierbaseline':
model_class = HierBaseline
backward_size = config.backward_size
else:
raise ValueError("This shouldn't happen.")
# LDA Model
ldamodel = LDAModel(config,
trained_model_path=FLAGS.lda_model_path,
id2word_path=FLAGS.id2word_path)
# get data set
api = SWDADialogCorpus(FLAGS.data_dir,
word2vec=FLAGS.word2vec_path,
word2vec_dim=config.embed_size,
vocab_dict_path=FLAGS.vocab_dict_path,
lda_model=ldamodel,
imdb=FLAGS.use_imdb)
dial_corpus = api.get_dialog_corpus()
meta_corpus = api.get_meta_corpus()
train_meta, valid_meta, test_meta = meta_corpus.get(
"train"), meta_corpus.get("valid"), meta_corpus.get("test")
train_dial, valid_dial, test_dial = dial_corpus.get(
"train"), dial_corpus.get("valid"), dial_corpus.get("test")
# convert to numeric input outputs that fits into TF models
train_feed = SWDADataLoader("Train", train_dial, train_meta, config)
valid_feed = SWDADataLoader("Valid", valid_dial, valid_meta, config)
test_feed = SWDADataLoader("Test", test_dial, test_meta, config)
# if you're testing an existing implementation or resuming training
if FLAGS.forward_only or FLAGS.resume:
log_dir = os.path.join(FLAGS.work_dir + "_" + FLAGS.model_type,
FLAGS.test_path)
else:
log_dir = os.path.join(FLAGS.work_dir + "_" + FLAGS.model_type,
"run" + str(int(time.time())))
# begin training
with tf.Session() as sess:
initializer = tf.random_uniform_initializer(-1.0 * config.init_w,
config.init_w)
scope = "model"
with tf.variable_scope(scope, reuse=None, initializer=initializer):
model = model_class(
sess,
config,
api,
log_dir=None if FLAGS.forward_only else log_dir,
forward=False,
scope=scope)
with tf.variable_scope(scope, reuse=True, initializer=initializer):
valid_model = model_class(sess,
valid_config,
api,
log_dir=None,
forward=False,
scope=scope)
with tf.variable_scope(scope, reuse=True, initializer=initializer):
test_model = model_class(sess,
test_config,
api,
log_dir=None,
forward=True,
scope=scope)
print("Created computation graphs")
if api.word2vec is not None and not FLAGS.forward_only:
print("Loaded word2vec")
sess.run(model.embedding.assign(np.array(api.word2vec)))
# write config to a file for logging
if not FLAGS.forward_only:
with open(os.path.join(log_dir, "run.log"), "wb") as f:
f.write(pp(config, output=False))
# create a folder by force
ckp_dir = os.path.join(log_dir, "checkpoints")
if not os.path.exists(ckp_dir):
os.mkdir(ckp_dir)
ckpt = tf.train.get_checkpoint_state(ckp_dir)
print("Created models with fresh parameters.")
sess.run(tf.global_variables_initializer())
if ckpt:
print("Reading dm models parameters from %s" %
ckpt.model_checkpoint_path)
model.saver.restore(sess, ckpt.model_checkpoint_path)
# if you're training a model
if not FLAGS.forward_only:
dm_checkpoint_path = os.path.join(
ckp_dir, model.__class__.__name__ + ".ckpt")
global_t = 1
patience = 10 # wait for at least 10 epoch before stop
dev_loss_threshold = np.inf
best_dev_loss = np.inf
# train for a max of max_epoch's. saves the model after the epoch if it's some amount better than current best
for epoch in range(config.max_epoch):
print(">> Epoch %d with lr %f" %
(epoch, model.learning_rate.eval()))
# begin training
if train_feed.num_batch is None or train_feed.ptr >= train_feed.num_batch:
train_feed.epoch_init(config.batch_size,
backward_size,
config.step_size,
shuffle=True)
global_t, train_loss = model.train(
global_t,
sess,
train_feed,
update_limit=config.update_limit)
# begin validation and testing
valid_feed.epoch_init(valid_config.batch_size,
valid_config.backward_size,
valid_config.step_size,
shuffle=False,
intra_shuffle=False)
valid_loss = valid_model.valid("ELBO_VALID", sess, valid_feed)
test_feed.epoch_init(test_config.batch_size,
test_config.backward_size,
test_config.step_size,
shuffle=True,
intra_shuffle=False)
test_model.test(
sess, test_feed, num_batch=1
) #TODO change this batch size back to a reasonably large number
done_epoch = epoch + 1
# only save a models if the dev loss is smaller
# Decrease learning rate if no improvement was seen over last 3 times.
if config.op == "sgd" and done_epoch > config.lr_hold:
sess.run(model.learning_rate_decay_op)
if True: #valid_loss < best_dev_loss: # TODO this change makes the model always save. Change this back when corpus not trivial
if True: #valid_loss <= dev_loss_threshold * config.improve_threshold:
patience = max(patience,
done_epoch * config.patient_increase)
# dev_loss_threshold = valid_loss
# still save the best train model
if FLAGS.save_model:
print("Save model!!")
model.saver.save(sess,
dm_checkpoint_path,
global_step=epoch)
# best_dev_loss = valid_loss
if config.early_stop and patience <= done_epoch:
print("!!Early stop due to run out of patience!!")
break
# print("Best validation loss %f" % best_dev_loss)
print("Done training")
# else if you're just testing an existing model
else:
# begin validation
valid_feed.epoch_init(valid_config.batch_size,
valid_config.backward_size,
valid_config.step_size,
shuffle=False,
intra_shuffle=False)
valid_model.valid("ELBO_VALID", sess, valid_feed)
test_feed.epoch_init(valid_config.batch_size,
valid_config.backward_size,
valid_config.step_size,
shuffle=False,
intra_shuffle=False)
valid_model.valid("ELBO_TEST", sess, test_feed)
# begin testing
dest_f = open(os.path.join(log_dir, "test.txt"), "wb")
test_feed.epoch_init(test_config.batch_size,
test_config.backward_size,
test_config.step_size,
shuffle=False,
intra_shuffle=False)
test_model.test(sess,
test_feed,
num_batch=None,
repeat=10,
dest=dest_f)
dest_f.close()
def main():
# config for training
config = Config()
# config for validation
valid_config = Config()
# valid_config.keep_prob = 1.0
# valid_config.dec_keep_prob = 1.0
# valid_config.batch_size = 60
# configuration for testing
test_config = Config()
test_config.keep_prob = 1.0
test_config.dec_keep_prob = 1.0
test_config.batch_size = 1
config.n_state = FLAGS.n_state
valid_config.n_state = FLAGS.n_state
test_config.n_state = FLAGS.n_state
config.with_direct_transition = FLAGS.with_direct_transition
valid_config.with_direct_transition = FLAGS.with_direct_transition
test_config.with_direct_transition = FLAGS.with_direct_transition
config.with_word_weights = FLAGS.with_word_weights
valid_config.with_word_weights = FLAGS.with_word_weights
test_config.with_word_weights = FLAGS.with_word_weights
pp(config)
print(config.n_state)
print(config.with_direct_transition)
print(config.with_word_weights)
# get data set
# api = SWDADialogCorpus(FLAGS.data_dir, word2vec=FLAGS.word2vec_path, word2vec_dim=config.embed_size)
with open(config.api_dir, "r") as fh:
api = pkl.load(fh)
dial_corpus = api.get_dialog_corpus()
if config.with_label_loss:
labeled_dial_labels = api.get_state_corpus(
config.max_dialog_len)['labeled']
# meta_corpus = api.get_meta_corpus()
# train_meta, valid_meta, test_meta = meta_corpus.get("train"), meta_corpus.get("valid"), meta_corpus.get("test")
train_dial, labeled_dial, test_dial = dial_corpus.get(
"train"), dial_corpus.get("labeled"), dial_corpus.get("test")
# convert to numeric input outputs that fits into TF models
train_feed = SWDADataLoader("Train", train_dial, config)
# valid_feed = SWDADataLoader("Valid", valid_dial, valid_meta, config)
test_feed = SWDADataLoader("Test", test_dial, config)
if config.with_label_loss:
labeled_feed = SWDADataLoader("Labeled",
labeled_dial,
config,
labeled=True)
valid_feed = test_feed
if FLAGS.forward_only or FLAGS.resume:
log_dir = os.path.join(FLAGS.work_dir, FLAGS.test_path)
else:
log_dir = os.path.join(FLAGS.work_dir, "run" + str(int(time.time())))
# begin training
with tf.Session(config=tf.ConfigProto(log_device_placement=True,
allow_soft_placement=True)) as sess:
initializer = tf.random_uniform_initializer(-1.0 * config.init_w,
config.init_w)
scope = "model"
with tf.variable_scope(scope, reuse=None, initializer=initializer):
model = VRNN(sess,
config,
api,
log_dir=None if FLAGS.forward_only else log_dir,
scope=scope)
with tf.variable_scope(scope, reuse=True, initializer=initializer):
valid_model = VRNN(sess,
valid_config,
api,
log_dir=None,
scope=scope)
#with tf.variable_scope(scope, reuse=True, initializer=initializer):
# test_model = KgRnnCVAE(sess, test_config, api, log_dir=None, forward=True, scope=scope)
print("Created computation graphs")
if api.word2vec is not None and not FLAGS.forward_only:
print("Loaded word2vec")
sess.run(model.W_embedding.assign(np.array(api.word2vec)))
# write config to a file for logging
if not FLAGS.forward_only:
with open(os.path.join(log_dir, "run.log"), "wb") as f:
f.write(pp(config, output=False))
# create a folder by force
ckp_dir = os.path.join(log_dir, "checkpoints")
if not os.path.exists(ckp_dir):
os.mkdir(ckp_dir)
ckpt = tf.train.get_checkpoint_state(ckp_dir)
print("Created models with fresh parameters.")
sess.run(tf.global_variables_initializer())
if ckpt:
print("Reading dm models parameters from %s" %
ckpt.model_checkpoint_path)
model.saver.restore(sess, ckpt.model_checkpoint_path)
#print([str(op.name) for op in tf.get_default_graph().get_operations()])
print([str(v.name) for v in tf.global_variables()])
import sys
# sys.exit()
if not FLAGS.forward_only:
dm_checkpoint_path = os.path.join(
ckp_dir, model.__class__.__name__ + ".ckpt")
global_t = 1
patience = config.n_epoch # wait for at least 10 epoch before stop
dev_loss_threshold = np.inf
best_dev_loss = np.inf
for epoch in range(config.max_epoch):
print(">> Epoch %d with lr %f" %
(epoch, model.learning_rate.eval()))
# begin training
if train_feed.num_batch is None or train_feed.ptr >= train_feed.num_batch:
train_feed.epoch_init(config.batch_size, shuffle=True)
if config.with_label_loss:
labeled_feed.epoch_init(len(labeled_dial), shuffle=False)
else:
labeled_feed = None
labeled_dial_labels = None
global_t, train_loss = model.train(
global_t,
sess,
train_feed,
labeled_feed,
labeled_dial_labels,
update_limit=config.update_limit)
# begin validation
valid_feed.epoch_init(config.batch_size, shuffle=False)
valid_loss = valid_model.valid("ELBO_VALID", sess, valid_feed,
labeled_feed,
labeled_dial_labels)
"""
test_feed.epoch_init(test_config.batch_size, test_config.backward_size,
test_config.step_size, shuffle=True, intra_shuffle=False)
test_model.test(sess, test_feed, num_batch=5)
"""
done_epoch = epoch + 1
# only save a models if the dev loss is smaller
# Decrease learning rate if no improvement was seen over last 3 times.
if config.op == "sgd" and done_epoch > config.lr_hold:
sess.run(model.learning_rate_decay_op)
"""
if valid_loss < best_dev_loss:
if valid_loss <= dev_loss_threshold * config.improve_threshold:
patience = max(patience, done_epoch * config.patient_increase)
dev_loss_threshold = valid_loss
# still save the best train model
if FLAGS.save_model:
print("Save model!!")
model.saver.save(sess, dm_checkpoint_path, global_step=epoch)
best_dev_loss = valid_loss
"""
# still save the best train model
if FLAGS.save_model:
print("Save model!!")
model.saver.save(sess,
dm_checkpoint_path,
global_step=epoch)
if config.early_stop and patience <= done_epoch:
print("!!Early stop due to run out of patience!!")
break
print("Best validation loss %f" % best_dev_loss)
print("Done training")
else:
# begin validation
# begin validation
global_t = 1
for epoch in range(1):
print("test-----------")
print(">> Epoch %d with lr %f" %
(epoch, model.learning_rate.eval()))
if not FLAGS.use_test_batch:
# begin training
if train_feed.num_batch is None or train_feed.ptr >= train_feed.num_batch:
train_feed.epoch_init(config.batch_size, shuffle=False)
if config.with_label_loss:
labeled_feed.epoch_init(len(labeled_dial), shuffle=False)
else:
labeled_feed = None
labeled_dial_labels = None
results, fetch_results = model.get_zt(
global_t,
sess,
train_feed,
update_limit=config.update_limit,
labeled_feed=labeled_feed,
labeled_labels=labeled_dial_labels)
with open(FLAGS.result_path, "w") as fh:
pkl.dump(results, fh)
with open(FLAGS.result_path + ".param.pkl", "w") as fh:
pkl.dump(fetch_results, fh)
else:
print("use_test_batch")
# begin training
valid_feed.epoch_init(config.batch_size, shuffle=False)
if config.with_label_loss:
labeled_feed.epoch_init(len(labeled_dial), shuffle=False)
else:
labeled_feed = None
labeled_dial_labels = None
results, fetch_results = model.get_zt(
global_t,
sess,
valid_feed,
update_limit=config.update_limit,
labeled_feed=labeled_feed,
labeled_labels=labeled_dial_labels)
with open(FLAGS.result_path, "w") as fh:
pkl.dump(results, fh)
with open(FLAGS.result_path + ".param.pkl", "w") as fh:
pkl.dump(fetch_results, fh)
"""
def main():
# config for training
config = Config()
print("Normal train config:")
# pp(config)
valid_config = Config()
valid_config.dropout = 0
valid_config.batch_size = 20
# config for test
test_config = Config()
test_config.dropout = 0
test_config.batch_size = 1
# LOG #
if not os.path.isdir('./output'):
os.makedirs('./output')
if not os.path.isdir('./output/{}'.format(args.expname)):
os.makedirs('./output/{}'.format(args.expname))
cur_time = str(datetime.now().strftime('%Y%m%d%H%M'))
# save arguments
json.dump(
vars(args),
open('./output/{}/{}_args.json'.format(args.expname, cur_time), 'w'))
logger = logging.getLogger(__name__)
logging.basicConfig(level=logging.DEBUG, format="%(message)s")
fh = logging.FileHandler("./output/{}/logs_{}.txt".format(
args.expname, cur_time))
# add the handlers to the logger
logger.addHandler(fh)
logger.info(vars(args))
###############################################################################
# Load data
###############################################################################
# sentiment data path: ../ final_data / poem_with_sentiment.txt
# 该path必须命令行显示输入LoadPoem，因为defaultNone
# 处理pretrain数据和完整诗歌数据
api = LoadPoem(args.train_data_dir, args.test_data_dir,
args.max_vocab_size)
# 交替训练，准备大数据集
poem_corpus = api.get_poem_corpus() # corpus for training and validation
test_data = api.get_test_corpus() # 测试数据
# 三个list，每个list中的每一个元素都是 [topic, last_sentence, current_sentence]
train_poem, valid_poem, test_poem = poem_corpus.get(
"train"), poem_corpus.get("valid"), test_data.get("test")
train_loader = SWDADataLoader("Train", train_poem, config)
valid_loader = SWDADataLoader("Valid", valid_poem, config)
test_loader = SWDADataLoader("Test", test_poem, config)
print("Finish Poem data loading, not pretraining or alignment test")
if not args.forward_only:
###############################################################################
# Define the models and word2vec weight
###############################################################################
# 处理用四库全书训练的word2vec
# if args.model != "Seq2Seq"
# logger.info("Start loading siku word2vec")
# pretrain_weight = None
# if os.path.exists(args.word2vec_path):
# pretrain_vec = {}
# word2vec = open(args.word2vec_path)
# pretrain_data = word2vec.read().split('\n')[1:]
# for data in pretrain_data:
# data = data.split(' ')
# pretrain_vec[data[0]] = [float(item) for item in data[1:-1]]
# # nparray (vocab_len, emb_dim)
# pretrain_weight = process_pretrain_vec(pretrain_vec, api.vocab)
# logger.info("Successfully loaded siku word2vec")
# import pdb
# pdb.set_trace()
# 无论是否pretrain，都使用高斯混合模型
# pretrain时，用特定数据训练特定的高斯分布
# 不用pretrain时，用大数据训练高斯混合分布
if args.model == "Seq2Seq":
model = Seq2Seq(config=config, api=api)
else:
model = PoemWAE(config=config, api=api)
if use_cuda:
model = model.cuda()
# if corpus.word2vec is not None and args.reload_from<0:
# print("Loaded word2vec")
# model.embedder.weight.data.copy_(torch.from_numpy(corpus.word2vec))
# model.embedder.weight.data[0].fill_(0)
###############################################################################
# Start training
###############################################################################
# model依然是PoemWAE_GMP保持不变，只不过，用这部分数据强制训练其中一个高斯先验分布
# pretrain = True
tb_writer = SummaryWriter(
"./output/{}/{}/{}/logs/".format(args.model, args.expname, args.dataset)\
+ datetime.now().strftime('%Y%m%d%H%M')) if args.visual else None
global_iter = 1
cur_best_score = {
'min_valid_loss': 100,
'min_global_itr': 0,
'min_epoch': 0,
'min_itr': 0
}
train_loader.epoch_init(config.batch_size, shuffle=True)
# model = load_model(3, 3)
batch_idx = 0
while global_iter < 100:
batch_idx = 0
while True: # loop through all batches in training data
# train一个batch
model, finish_train, loss_records = \
train_process(model=model, train_loader=train_loader, config=config, sentiment_data=False)
batch_idx += 1
if finish_train:
test_process(model=model,
test_loader=test_loader,
test_config=test_config,
logger=logger)
evaluate_process(model=model,
valid_loader=valid_loader,
global_iter=global_iter,
epoch=global_iter,
logger=logger,
tb_writer=tb_writer,
api=api)
# save model after each epoch
save_model(model=model,
epoch=global_iter,
global_iter=global_iter,
batch_idx=batch_idx)
logger.info(
'Finish epoch %d, current min valid loss: %.4f \
correspond global_itr: %d epoch: %d itr: %d \n\n' %
(global_iter, cur_best_score['min_valid_loss'],
cur_best_score['min_global_itr'],
cur_best_score['min_epoch'],
cur_best_score['min_itr']))
# 初始化下一个unlabeled data epoch的训练
# unlabeled_epoch += 1
train_loader.epoch_init(config.batch_size, shuffle=True)
break
# elif batch_idx >= start_batch + config.n_batch_every_iter:
# print("Finish unlabel epoch %d batch %d to %d" %
# (unlabeled_epoch, start_batch, start_batch + config.n_batch_every_iter))
# start_batch += config.n_batch_every_iter
# break
# 写一下log
if batch_idx % (train_loader.num_batch // 50) == 0:
log = 'Global iter %d: step: %d/%d: ' \
% (global_iter, batch_idx, train_loader.num_batch)
for loss_name, loss_value in loss_records:
log = log + loss_name + ':%.4f ' % loss_value
if args.visual:
tb_writer.add_scalar(loss_name, loss_value,
global_iter)
logger.info(log)
# valid
if batch_idx % (train_loader.num_batch // 10) == 0:
valid_process(
model=model,
valid_loader=valid_loader,
valid_config=valid_config,
global_iter=global_iter,
unlabeled_epoch=
global_iter, # 如果sample_rate_unlabeled不是1，这里要在最后加一个1
batch_idx=batch_idx,
tb_writer=tb_writer,
logger=logger,
cur_best_score=cur_best_score)
test_process(model=model,
test_loader=test_loader,
test_config=test_config,
logger=logger)
save_model(model=model,
epoch=global_iter,
global_iter=global_iter,
batch_idx=batch_idx)
# if batch_idx % (train_loader.num_batch // 3) == 0:
# test_process(model=model, test_loader=test_loader, test_config=test_config, logger=logger)
global_iter += 1
# forward_only 测试
else:
# test_global_list = [4, 4, 2]
# test_epoch_list = [21, 19, 8]
test_global_list = [8]
test_epoch_list = [20]
for i in range(1):
# import pdb
# pdb.set_trace()
model = load_model('./output/basic/header_model.pckl')
model.vocab = api.vocab
model.rev_vocab = api.rev_vocab
test_loader.epoch_init(test_config.batch_size, shuffle=False)
last_title = None
while True:
model.eval() # eval()主要影响BatchNorm, dropout等操作
batch = get_user_input(api.rev_vocab, config.title_size)
# batch = test_loader.next_batch_test() # test data使用专门的batch
# import pdb
# pdb.set_trace()
if batch is None:
break
title_list, headers, title = batch # batch size是1，一个batch写一首诗
if title == last_title:
continue
last_title = title
title_tensor = to_tensor(title_list)
# test函数将当前batch对应的这首诗decode出来，记住每次decode的输入context是上一次的结果
output_poem = model.test(title_tensor=title_tensor,
title_words=title_list,
headers=headers)
with open('./content_from_remote.txt', 'w') as file:
file.write(output_poem)
print(output_poem)
print('\n')
print("Done testing")
